6

I’m working on a paper and I realized I do not have an adequate dataset (for reference, this is in a subfield of machine learning), nor do I have the time to personally go and create one. A colleague and friend of mine offered to collect the data and give it to me, as it would be useful as a benchmark for some of his own experiments later.

In this case, “collecting data” means finding and equipping participants with technology to monitor them perform a simple task for a brief amount of time (5 min).

Since this colleague is a personal friend whom I’ve worked with in the past, I have no qualms about listing him as a coauthor (and will most certainly do so, unless it is unethical for some reason). I am curious as to the general case of whether providing someone with an unpublished dataset is insufficient, sufficient or should guarantee coauthorship. I should emphasise that in my field, data is not something that begets the results but rather serves as a test to determine if a proposed technique or algorithm is well-suited to the task outlined by the data.

To put it simply: is providing data enough of a contribution to warrant coauthorship? Would it be unethical to list someone as a coauthor if they only created the dataset?

  • 1
    Just a note: You're not going to get a yes/no answer to this question (and if you do, you should discount it because this isn't a yes/no situation.) The best you can hope for is a guideline that you can apply to your own situation to judge for yourself. – ff524 Dec 13 '15 at 3:42
  • 1
    I'm really hoping to foster interesting and relevant discussion more than get a hard answer. I fully expect that the only generalized answer is "Well, maybe." I just realized that this question could help/inform others in the future with similar situations. – ahjohnston25 Dec 13 '15 at 3:51
  • 2
    Vote to close. This isn't really the place to "foster interesting discussion." – Corvus Dec 13 '15 at 5:22
  • 3
    @Corvus: I fail to see how this is different from many other authorship questions we get. It’s a clear, yet not overly localised scenario. The only possible problem I see is that the answer may be field-dependent, but I have no background in such fields that would allow me to say. – Wrzlprmft Dec 13 '15 at 7:33
  • 2
    When in doubt, offer co-authorship. It is then up to them to decline or accept. – Alexandros Dec 13 '15 at 8:22
7

The Vancouver protocol gives authorship according to the following criteria:

Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; AND

Drafting the work or revising it critically for important intellectual content; AND

Final approval of the version to be published; AND

Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

So collecting data fulfills the first criteria. Therefore, the nice thing to do is to offer a co-authorship for your colleague. It is then up to him or her to decide whether he or she would like participate, which would entail fulfilling the other three criteria as well.

In this case, “collecting data” means finding and equipping participants with technology to monitor them perform a simple task for a brief amount of time (5 min).

Be careful not to underestimate the amount of time needed to recruit participants. The recruiter usually has to write several emails and advertisements, answer phone calls, brief and debrief the participants, prepare and take written informed consents, etc.

In answer to the comment, if your colleague accepts the co-authorship, ask him or her to write about the data collection, participants, motivation and possible limitations of the test, how the results can be interpreted and so on. Machine learning that matters should communicate something back to problem domain. This would be sufficient to fulfill the second Vancouver criteria.

Anyway, some fields have very different conventions, so your mileage may vary.

2

Obviously standards vary across fields. In some physical sciences, providing the data might not rise to authorship if the process of obtaining them was not considered significant creative work. Something similar would hold in many of the social sciences unless the data were really something special. In biology, my understanding is that providing new unpublished data would certainly warrant coauthorship and that providing old published data is often enough to clear the bar.

  • 1
    That's why I tried to explain a bit about my field. In some fields (such as biology), data is essentially unprocessed results. In mine, it is merely how we find out if a proposed algorithm or technique is suitable for a specific task. Thousands of people can all work on the same dataset and produce different and publishable results. – ahjohnston25 Dec 13 '15 at 2:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.